Convfit problems

We have been trying to use Convfit with our QENS spectra from Osiris, and some spectra give an excellent fit, but others seem to fall into false minima that are obviously wrong. In some cases this can be overcome by adjusting the parameters by hand, but presumably it shouldn’t be necessary to do this spectrum by spectrum.
The documentation says that it is possible to tie together certain parameters for different spectra (for example in our case we think there is no translational motion, and so we believe the FWHM should be constant for all Q), but it is not clear from the documentation how to do this on IDAAS, because the specific manual instructions do not entirely match up with what happens on IDAAS. Assuming “L” and “G” settings are local and global, they both still seem to be available regardless of whether a simultaneous or sequential fit is chosen. If “G” is chosen, there is no visible option to use a tie. If “L” is chosen, what is the syntax for tying a parameter in a different spectrum?

It might not be clear, but there are three dots that become visible when you move the mouse near the fitting parameter. When you click on it, a dialogue with tying options appears